How Tencent Turns Game Operations into Service‑Driven Success

Tencent Game Operations Service Construction Journey

With the rise of big data and cloud computing, traditional operations can no longer meet the demands for user experience and efficiency, especially in game operations.

At Tencent, game operations not only handle daily releases, changes, incidents, and migrations, but also must identify business and user pain points and drive solutions through technology, which forms the core value of operations.

Operations now go beyond building backend infrastructure; they are responsible for game operability, experience optimization, real‑time data statistics, and architecture improvements.

The transition from traditional IT operations to business‑oriented technical operations requires clear direction, long‑term practice, robust tooling platforms, and a comprehensive service system.

Overall, Tencent's game service construction can be divided into three stages:

1. Proposal of the Four Transformations

Before 2012, the Tencent Interactive Entertainment Operations team introduced the four transformations for operations: standardization, automation, specialization, and service orientation.

2. Exploration of an Operations Service System (2013)

With the growth of underlying tool platforms, the team began building value‑added services aimed at "reaching users and improving direct business benefits," such as user‑reach optimization and high‑availability & cost‑optimization services.

The early service framework looked like this:

3. Formal Launch of the Tencent Game Service System (2014)

As service instances accumulated, automation of basic operations increased, clarifying the core value: "solve business and user pain points through technical solutions and provide a measurable service framework."

The system consists of six modules:

User Experience Optimization Service

Operation Activity Service

Version Service

Operation Cost Control & Optimization Service

Business Security Assurance Service

Operations Consulting Service

Case 1: User Experience Optimization from Pre‑Game Entry

Players must go through download, installation, upgrade, and login before entering a client game; any step can become a barrier.

Business data insight: Over half of players abandon downloads due to slow speed.

Technical approach: Targeted acceleration rather than blanket speed increases, focusing on IP‑based and QQ‑number‑based acceleration.

IP‑based acceleration: Identify regions with low download completion rates and boost speed selectively.

QQ‑number acceleration: Accelerate specific account groups during acquisition or re‑engagement campaigns.

Further solutions include per‑user speed allocation and collaborating with ISPs to break home‑bandwidth limits.

Dynamic intelligent throttling adjusts download speeds before peak bandwidth periods, reducing concurrent peaks and keeping bandwidth usage stable.

These measures raised the average download success rate by 15% and player conversion by nearly 10%.

Case 2: Game Live Streaming

Live streaming in games demands high network and hardware resources; Tencent’s QQ Dance launched star live streams in 2013, using a transcoding pipeline that creates two modes: normal streamer and concert mode.

Business Issues and Architecture

Traffic: Large traffic handled by deploying access points in multiple IDC locations, using ping‑based random ordering and dedicated points for large events.

Resource sharing: A unified live‑room architecture enables sharing and horizontal scaling, reducing operational costs.

User Experience – Video Lag

Analysis revealed two key factors beyond network quality: retransmission mechanisms and uplink bandwidth limits on the streamer side.

When retransmission occurs, uplink traffic can spike from 20‑30 KB/s to over 220 KB/s, causing congestion and a vicious cycle of lag.

Technical solutions:

Optimize retransmission to reduce attempts and uplink load.

Enable uplink bandwidth limits for users.

Coordinate with PC manager to disable unnecessary uploads from music or download software.

Continuous Operational Monitoring

After improvements, the system monitors network fluctuations and CDN anomalies, providing real‑time alerts and self‑help suggestions to streamers.

Current CDN connection and processing success rates exceed 95%.

Case 3: Server Merge Decision Support

Server merging combines isolated game zones to boost activity and lower costs, but simple script execution is insufficient for business‑oriented outcomes.

The operations team focuses on four aspects:

Merge decision factors: Multi‑dimensional data integration allows product teams to define custom conditions for selecting merge zones.

Automated execution: One‑click trigger completes the merge process.

Data tracking: System outputs pre‑ and post‑merge performance comparisons.

Decision assistance: Historical merge results are analyzed to recommend optimal future merge conditions.

The system automatically calculates eligible merge zones, provides product teams with options, and generates effect reports after each merge.

By analyzing merge outcomes against historical data, the platform regularly recommends the most profitable merge candidates.

Conclusion

Through these operational case studies, the team demonstrates a shift from passive support to proactive, business‑driven technology operations, continuously refining techniques, aligning closely with business needs, and leading intelligent game operation.